Rapid diagnostic tests (RDTs), specifically immunochromatographic lateral flow assays, can provide accurate real-time point-of-care diagnoses in low-resource settings and have been an important tool in the global health arsenal. Advances in microfluidics have enabled the medical device community to design smaller RDTs—as small as half the area of a business card and the thickness of a watch—that can be used to diagnose more conditions and are low cost—less than US $1 per device [1]. Manufacturers and international health organizations have collaborated to deliver hundreds of millions of RDTs to countries and communities [2]. However, the full potential of RDTs as a public health tool has not yet been realized. This is due to field deployment challenges amplified by a fragmented market, nonstandard designs, and lack of features that facilitate systematic capture and use of RDT results and patient data. Fortunately, current technology—computer vision, widely deployed smartphones, and mobile networks—applied to RDTs provides a scalable path to improved health and well-being through the emerging paradigm of precision public health (PPH). PPH is a field that aims to maximize impact with the active use of data for surveillance and targeted interventions by temporal, spatial, and epidemiological characteristics of populations.

Therefore, we aimed to define axioms to underpin steps to incorporate RDTs into a PPH approach, identify an initial set of features (RDT’s hardware and software features) that would be needed to implement these axioms, and select a final evidence-based set of features that could be used by health policy and program implementers in integrating RDTs as tools for frontline health care workers at the community and clinic levels [3,4].

There are 3 core challenges faced within the current RDT ecosystem, which impede application and widespread implementation for PPH.

Based on these challenges, we define 3 axioms that underpin solutions (Table 1). To solve the challenges and maximize information usage for PPH, RDTs should adhere to a set of open guidelines (OGs), and use smartphone readers and data protocols that standardize the information—both horizontally between different manufacturers or providers and vertically between different steps in the RDT life cycle (Figure 1). These axioms can transform the current state of the RDT ecosystem into one that supports PPH. RDT-OGs address the data uniformity challenge of RDTs by standardizing the capture and use of data.

We have chosen to address the current challenges in RDTs with open guidelines in order to focus directly on the systems and integration problems they face. Figure 2 shows initial progress as RDTs were developed in the laboratory; as researchers predicted their impact, optimism surrounding their potential grew. As RDT rollouts began, the ability to capture diagnostic data increased, but these increases did not keep pace with global growth in the technological capacity to store, communicate, and analyze information [9].

This changing technology landscape, combined with a lack of individual RDT identifiers, inconsistent test use protocols, and the appearance of fraudulent and counterfeit RDTs, led to a relative decrease in information use; however, as the RDT community began to effectively encode and aggregate information and improve the training of health care workers, information use increased. Currently, RDTs have reached a tipping point—there are multiple proprietary hardware and software solutions, and medical systems are facing “information overload” [10]. The future trend in information use could take 1 of 2 diverging paths—modest growth with eventual stagnation or a promising future with open guidelines aligned with the aforementioned PPH axioms to accelerate impact by enhancing information usage (Figure 2).

Below, we describe our methods, present survey results from experts in RDT technology, formally establish an information utilization index (IUI), and define how various RDT features gather information. We use this to assess World Health Organization (WHO)–prequalified RDTs in comparison to a prototype RDT based on open guidelines and then discuss these results and related work in field-based hardware and software diagnostics and standards. As access to telecommunications networks improves worldwide, and advanced information systems become more widely used by ministries of health and global health organizations, the community can dramatically accelerate the transformational impact promised by RDTs.

To identify key issues related to data capture and use from RDTs, we conducted a review of published papers using the Semantic Scholar artificial intelligence–enabled research search engine [11]. We focused on, but did not constrain ourselves to, PubMed-indexed medical journal papers. The search was conducted on December 31, 2018, and updated on December 31, 2019, using the keywords “rdt,” “smartphone,” and “mobile phone.” The search revealed 480 papers that were further screened for the study of lateral flow immunochromatographic rapid tests, yielding 58 papers. In addition to reviewing these papers, we reviewed their citations to identify additional papers in telecytology, immunochromatography, and diagnostic hardware. From these papers, we extracted themes and concepts related to barriers to information capture and usage from RDTs, and applied a grounded theory conceptual framework to compile and code themes and concepts into core ideas and then high-level abstractions and classifications. These were discussed and reviewed by 3 different members of the team. We considered this process complete when saturation was reached (ie, no additional novel ideas or abstractions emerged upon review of additional papers).

Survey results and the literature review were used to identify essential information features for RDTs and their integration into health care platforms to support PPH. The presence or absence of a feature for a specific RDT could be used to calculate an IUI defined as number of features present or the number of features defined. We then selected WHO prequalified cassette-based RDTs for malaria and HIV that had been assessed for performance [12] and calculated the IUI.

The review of published literature and thematic extraction of concepts related to information capture and usage from RDTs led us to identify the following core areas that affect information usage (Table 2).

We contacted 81 stakeholders, and 33 completed the questionnaire (16 researchers, 7 technologists, 3 manufacturers, 2 doctors or nurses, and 5 others). Respondents were most concerned about the proper use of RDTs (agreed: 30/33, 91%), their interpretation (agreed: 28/33, 85%), and reliability (agreed: 26/33, 79%). Respondents were confident that smartphone-based RDT readers could address some reliability concerns (agreed: 28/33, 85%) and that readers were more important for complex or multiplex RDTs (agreed: 33/33, 100%).

Based on these results, and because RDTs are embedded in a set of protocols and practices defined by health care workers, institutions, clients, and communities, proper usage of rapid diagnostic tests depends not only on a physical device but also on its integration into the larger ecosystem. In this context, to maximize information usage, an RDT platform must function effectively in all phases of its life cycle, with added value at each phase.

Specific stakeholder results are divided into discrete phases of the rapid diagnostic test life cycle (Manufacture, Shipping, Use, Interpretation, and Disposal), and RDT capabilities are divided into themes (Metadata, Molding of the Cassette, Printed Data, and Smartphone Reader) (Figure 3). The open guidelines contribute to each theme, and they are essential for the Smartphone Reader theme. A conceptual framework that contrasts the accumulated value of RDT open guidelines and the current RDT process shows an increase at each life cycle phase. Specific capabilities drive these increases (Figure 3).

We identified 11 essential information features for RDTs and their integration into health care platforms, which define components of the IUI (Textbox 1).

The IUI—which provides an overview of how much information current diagnostics can capture and where there is room for improvement—for prequalified RDTs and an OG RDT had values ranging from 0 to 0.75 (mean 0.27; median 0.30, IQR 0.25) (Figure 4). The large bracket shows that 70% of this information usage score can be attributed to printed or other nonphysical changes, while the remaining 30% require physical changes to the RDT.

Our RDT-OGs recommend the horizontal integration of RDT hardware through consistent physical modules, thereby enabling vertical integration of RDT software through consistent protocols linking supply chain, test choreography, and interpretation. In Figure 5, a reference example of RDT-OGs with colored overlays identifying the core modifications is shown. These include a 2D barcode to embed information needed for an app to identify, read, and interpret the RDT; fiducials as reference points to assist the camera and phone to quickly and accurately identify the RDT areas of interest and reference; and a color calibration panel to enable reliable colorimetric inference. In addition, 3 WHO-prequalified RDTs with overlays are shown to highlight current inconsistencies between tests (Figure 5).

In comparing the optimized IUI and design of the reference RDT-OGs to others, we observed both the heterogeneity and common structures across all RDTs. We have designed RDT-OGs to be useful whether adopting all recommendations, a subset of modules, or using existing cassettes linked to an RDT-OG–compatible software platform. Defining common data models and schemas provides an information architecture that would encapsulate data from any module combination that exists on the RDT. The RDT-OG data schema can be effectively encoded by the 2D barcode and easily drive the process forward via a reader app. The design and production aspects have proved feasible given the successful production of the prototype, and the field assessments, such as assessment of integration with epidemiological monitoring systems, are ongoing.

Creating a systematic way (Figure 6) to collect and aggregate structured RDT data allows the community to continuously monitor device performance, disease prevalence, and the relationship between demographic priors and diagnostic outcomes. This workflow, like the RDT-OG, is not tied to a particular diagnostic and is designed to accommodate both existing and emerging diagnostics. RDTs can increase their IUI by using a universal RDT-OG–compatible reader and data storage. New RDTs that become available in the market can further increase their IUI by using the hardware recommendations of the RDT-OGs (Figure 6). The workflow integrates automatic result interpretation modules using machine learning from image libraries or template-based approaches and can dynamically accommodate new RDTs through database-backed parameters defining rapid diagnostic test components and hyperparameters identifying the specific RDT (Figure 7). The rapidly growing number of COVID-19 serology and antigen-based RDTs show the critical role of dynamically supporting newly released RDTs [13,14].

To the best of our knowledge, this is the first paper to propose guidelines to harmonize the hardware, software, and data standard used to read and interpret RDTs.

Existing standards for diagnostics encompass RDTs, including several related standards in health technology, medical devices, and precision medicine. For example, given that most rapid diagnostic tests have maximal temperature limits for storage and use, temperature exposure monitors such as those used on vaccine vials would be warranted based on performance degradation from heat exposure [34]. In addition, use of the Fast Healthcare Interoperability Resources (FHIR) [35] can define clinical data as a graph of well-defined fields and data types (Figure 7). A number of current digital lab data platforms and application programming interfaces already handle related diagnostic and laboratory information management with FHIR as a common standard [36].

Regulators, such as the US Food and Drug Administration, define pathways to classify novel medical devices, including communication-enabled RDTs, for which there are no similar existing devices [37]. Similarly, the Medical Device Communications Testing Project from the National Institute of Standards and Technology is relevant to any form of medical device communication and applicable to RDTs which communicate via radio frequency or electrochemical means [38]. These regulatory bodies also promote innovation (eg, the National Institute of Standards and Technology Text Retrieval Conference, where annual precision medicine competitions model the most effective treatments) exemplifying how communities can benefit from well-structured data [39].

This study has some limitations. First, the response rate from the survey was 40% (33/81), and the survey results may have benefited from additional feedback from a broader group. Nevertheless, there was strong thematic concordance between core responses from the survey and findings from the literature review. Second, after designing the open guidelines, we did not solicit additional feedback from the same and similar groups of persons who were contacted for the survey. This additional step would serve to validate the utility of the open guidelines. We note that our goal was to collect feedback from RDT producers and users of RDT-OGs. Third, we limited ourselves to information usage issues and solutions for cassette-based rapid tests and did not include the simpler dipstick type strips. However, the same concepts would apply and would need to be implemented in a way that is compatible with the lower space and cost profile of such tests [40]. Despite these caveats, the proposed RDT-OG approach is clearly applicable to the majority of RDTs currently deployed globally, and to those likely to be produced in the future as multiplex and complex tests become the norm.

In response to the goals and ambitions of the RDT community, we defined PPH axioms and derived RDT-OGs. The recommended modular foundation is designed to accelerate current RDT development, fieldwork, and successfully translate RDTs into effective field evaluations and deployments at scale. These guidelines thus confer functionality to diagnostic devices, the smartphone apps interpreting them, and the health information system analyzing them. For example, temperature sensors may be essential to assure proper storage and quality of some rapid diagnostic tests [34], and the modularity of open guidelines can accommodate this need. Although modifying supply chains may be infeasible in areas with rigid logistics or fixed asset costs, the RDT-OGs gives the community a pathway to extend the functionality of pre-existing field-based diagnostics through advances in machine learning, which do not require RDT modifications.

National and global policy makers have shown a willingness and ability to convene communities around guidelines that benefit RDT stakeholders; for example, the WHO prequalification of medicines program, FHIR, SNOMED, and LOINC. As the WHO, the Global Fund, Foundation for Innovative New Diagnostics, and others continue this work, there is ample opportunity to adopt formal guidelines around RDTs and their usage. For example, the WHO’s role in creating and promoting prequalified malaria RDTs has incentivized manufacturers to increase low-cost RDT production [41,42]. A similar approach to incentivize machine-readable RDT identifiers, and data schemas to interpret them, would likely address challenges currently faced.

Thus, by providing guidance for RDT hardware, software, and data interoperability, standards-setting organizations can transform RDTs into a formidable public health tool for disease prevention and treatment, in addition to diagnosis. These innovations can accelerate long-term disease control efforts, such as for malaria, which is responsible for 7.8% of the annual deaths of children under 5 years old (20,000 children worldwide [43]). Furthermore, these innovations can accelerate rapidly evolving disease control efforts, such as for COVID-19, where serological or antigen detection investigations face challenges in obtaining case information; these challenges are expected to further increase as testing efforts continue to scale up, and with transition from mitigation to containment [44]. Therefore, in both routine and emergency scenarios, adopting RDT-OGs would apply key advances in information technology to close the critical gap between diagnostics and public health interventions, and enable a new era of precision public health.